L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Développement · SKILL.md

testdino-manual-runs

testdino-manual-runs

Use when the user wants to manage a manual execution run or update case-level results inside a run — listing runs, creating runs for a release, inspecting a run, assigning cases, or marking case results (passed/failed/blocked/skipped/retest/untested). Accepts counter-style IDs like RUN-12 and TC-156.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: testdino-manual-runs description: Use when the user wants to manage a manual execution run or update case-level results inside a run — listing runs, creating runs for a release, inspecting a run, assigning cases, or marking case results (passed/failed/blocked/skipped/retest/untested). Accepts counter-style IDs like RUN-12 and TC-156.

TestDino Manual Runs

Use TestDino manual-run tools when the user wants to manage a manual execution run or update case-level results inside a run.

Preferred flow:

  1. Call health if projectId is not known.
  2. Use list_manual_runs to discover runs by name, state, release, environment, or tags.
  3. Use get_manual_run to inspect one run's metadata and rollup stats.
  4. Use create_manual_run or update_manual_run for run-level metadata changes.
  5. Use list_run_test_cases before changing any individual case in a run.
  6. Use update_run_test_case to assign a case, change its result, or update elapsed time.

Important details:

  • runId accepts an internal ID or a counter-style ID like RUN-12.
  • rtcRef can be the run test case row ID, case key like TC-156, or the underlying test case ID.
  • updates.assigneeUserId can be a user ID or email address.
  • updates.result accepts display form or canonical form, but canonical values like passed, failed, blocked, skipped, retest, and untested are safest.
  • For create and update payloads, tags and linkedIssues must be JSON arrays.

Good requests for this skill:

  • list manual runs for a release
  • create a manual run for sprint MS-12
  • show all test cases in run RUN-12
  • assign TC-156 in RUN-12 to alice@example.com
  • mark TC-156 as passed in RUN-12
Skills enregistrées